Thermal Radiation Modeling WorkshopFor SPIE Optics+Photonics 2017
Zhiguang Zhou1, Enas Sakr1, Tianran Liu2, Wendy Lin3, Evan Schlenker1, Hao Tian1, Cindy Karina4, Peter Bermel1
1 Birck Nanotechnology Center, Purdue University, West Lafayette, IN, USA2 Electrical Engineering, Princeton University, Princeton, NJ, USA3 Electrical Engineering and Computer Science, Georgia Institute of Technology, Atlanta, GA, USA4 Swiss Federal Institute of Technology (ETH-Zurich), Zurich, Switzerland
Brief History of Thermal Energy Harvesting
Year Discovery Significance Innovators
1712 Atmospheric Engine Pumping water out of mines Thomas Newcomen
1776 Steam Engine Mechanical workhorse of industrial revolution James Watt
1870 Kirchoff’s law of thermal
radiation
Establishing centrality of blackbody in thermal radiation Gustav Kirchoff
1879 Stefan-Boltzmann law Calculating total radiated power Josef Stefan
1900 Planck’s law of blackbody
radiation
Calculating radiation power spectrum Max Planck
1956 Thermophotovoltaics Converting thermal radiation into electricity Henry Kolm
1960 Laser Provides intense, monochromatic optical power Schawlow & Townes
1962 Solar Cell Efficiency Limits Provided a target for PV and TPV research Shockley & Queisser
1979 Gallium antimonide cell Provides suitable bandgap for TPV Lew Fraas
2014 Photonic radiative cooling Provides nearly ideal radiative cooling Shanhui Fan
SPIE Optics+Photonics, San Diego, CA - Peter Bermel2 August 6, 2017
Key Concepts from Prior Research
Carnot efficiency of heat engines
Planck blackbody limit: centrality of blackbody in thermal
radiation
Shockley-Queisser limit of photovoltaics
Additional losses at every step in practice
SPIE Optics+Photonics, San Diego, CA - Peter Bermel3 August 6, 2017
SPIE Optics+Photonics, San Diego, CA - Peter Bermel4 August 6, 2017
61% of raw energy wasted in 2013!
Energy Landscape Today
61% of raw energy wasted in 2013!
Energy Landscape Today
SPIE Optics+Photonics, San Diego, CA - Peter Bermel5 August 6, 2017
Too much parasitic loss in
commonly used devices,
like ovens and light bulbs
The chart above lists values of overall luminous efficacy and efficiency for several types of
general service, 120-volt, 1000-hour lifespan incandescent bulb
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel6
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel7
Make Solar Energy Economical
Key Challenges:
Novel earth-abundant materials
Reliable, low-cost packaging techniques
Energy storage (daily and seasonal)
How simulations can help:
Provide predictions of performance of realistic, novel PV materials
(e.g., using DFT)
Predict and optimize lifetime energy production (e.g., using ADEPT)
Design electrolyzers and fuel cells (e.g., using FEM multi-physics)
8
Lewis, N.S. 2007. Toward Cost-Effective Solar Energy Use. Science 315(5813): 798-801. DOI: 10.1126/science.1137014
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Wireless ESSID: Left-hand side: UNITE-9980; password: 78841109
Right-hand side: 791L-8337; password: 68e8dc6d
Nanohub login: https://nanohub.org/
Create account via ‘Signup’ link in upper right
Login with institutional login, Facebook, or LinkedIn
Bug reporting site
• https://nanohub.org/
• (upper right) help link
Get the hands-on files
– https://nanohub.org/groups/photonics/thermal_rad_workshop
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel 9
Step 1: Reach Maximum Temperature
from Solar Heat
Key tool(s):
S4sim
SPIE Optics+Photonics, San Diego, CA - Peter Bermel10 August 6, 2017
Selective Absorber:
Maximum Thermal Transfer Efficiency
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel11
P. Bermel et al., Ann. Rev. Heat Transfer (2012).
Thermal Transfer Efficiency
Spectrally-averaged absorptivity
Spectrally-averaged emissivity
Best Commercial Selective Solar
Absorbers: T=400 K (1 sun)
Almeco-TiNOX Solar
ht = 90%; a = 95%; e = 5%
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel12
http://www.almecogroup.com/en/pagina/16-solar
Photonic Simulations with
S4
Full-wave photonic simulations of arbitrary
layered media, including thin-film and
crystalline PV cells
V. Liu, S. Fan, Comp. Phys.
Comm. 183, 2233 (2012)
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter
Bermel
13
https://nanohub.org/tools/s4sim/
What is nanoHUB and S4?
An open-access science
gateway for cloud-based
simulation tools and resources
in nanoscale science and
technology.
Stanford Stratified Structure
Solver (S4) is a frequency
domain code to solve layered
periodic structures.
An input control file scripted in
LUA outputs Absorption
Spectrum using S-matrix Method
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel14
S-Matrix Method: Advantages
No ad hoc assumptions regarding structures
Applicable to wide variety of problems
Suitable for eigenmodes or high-Q resonant modes at single
frequency
Can treat layers with large difference in length scales
Computationally tractable enough on single core machines
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel15
S-Matrix Method: Disadvantages
Accurate solutions obtained more slowly as the following
increase:
Number of layers
Absolute magnitude of Fourier components (especially for metals)
Number of plane-wave components (~N3)
Relatively slow for broad-band problems (time-domain is a
good alternative)
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel16
Accuracy improves systematically with computing power
V. Liu, S. Fan, Comp. Phys. Comm. 183, 2233 (2012)
Photonic Simulations with S4
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel 17
S4: Input
Can choose several pre-made examples drawn from the literature
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel18
S4sim: Output Window
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel19
S4: Output
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Transmission through multilayer stack matches analytical expression
20
S4: Output
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Transmission through 1D square grating of silicon and air
21
S4: Output
Transmission from Fig. 4 of Tikhodeev et al.,
Phys. Rev. B 66, 045102 (2002).August 6, 2017
SPIE Optics+Photonics, San Diego, CA - Peter Bermel22
S4: Lua Control Files
Obtain a new, blank simulation object with no solutions:
S = S4.NewSimulation()
Define all materials:
S:AddMaterial("Dielectric", {4,0.1}) -- real and imag partsS:AddMaterial("ARmat", {2,0})S:AddMaterial("Vacuum", {1,0})
Add all layers:
S:AddLayer('AirAbove’,0,'Vacuum’)S:AddLayer('ARcoat', art, 'ARmat')S:AddLayer('Slab', 2, 'Dielectric')S:AddLayerCopy('AirBelow’,0,’AirAbove’)
Add patterning to layers:
S:SetLayerPatternCircle('layer_name', 'inside_material’, {0.0,0.0}, 0.2) –- centerx, centery, and radius
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel23
S4: FMM Formulations Specify the excitation mechanism:
S:SetExcitationPlanewave(
{0,0}, -- phi and theta: phi in [0,180), theta in [0,360)
{1.0,0.0}, -- s_pol_amp, s_pol_phase in degrees
{1.0,0.0}) -- p_pol_amp, p_pol_phase in degrees
S:SetNumG(1)
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel24
S4: FMM Formulations Specify the operating frequency range and output:
for freq=0.2,0.4,0.005 do
S:SetFrequency(freq)
forward_power, backward_power = S:GetPoyntingFlux(‘AirAbove', 0)
arf,arb = S:GetPoyntingFlux('ARcoat',0)
slab_forward,slab_backward = S:GetPoyntingFlux('Slab', 0)
E2 = S:GetLayerElectricEnergyDensityIntegral('Slab');
absorption = 1.0 - (math.abs(forward)+math.abs(abb))/math.abs(abf)
avea=avea+absorption*(freq-oldfreq)
denom=denom+freq-oldfreq
oldfreq=freq
end
avea=avea/denom
print(art .. ‘\t' .. avea);August 6, 2017
SPIE Optics+Photonics, San Diego, CA - Peter Bermel25
Results
SPIE Optics+Photonics, San Diego, CA - Peter Bermel26 August 6, 2017
S4sim Example: PV Front Coating
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Number of front coating
layers1 2 3
Relative permittivity Real Imag Real Imag Real Imag
Layer 1
Layer 2
Layer 3
4.32 0 2.37
9.12
0
0
1.80
5.71
14.36
0
0
0
Number of front coating
layers1 2 3
Thickness (nm)
Layer 1
Layer 2
Layer 3
60 82.3
38.9
91.0
53.1
29.9
27
S4sim Example: PV Front Coating
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Results from M. Ghebrebrhan, P. Bermel, Y. Avniel,
J. Joannopoulos, and S. Johnson, Optics Express
17, 7505-7518 (2009).
Results generated by S4sim
28
Selective Solar Absorbers
Si
Si3N4
Ag
215 nm
300 nm
300 um
Schematic of the structure for selective absorber based on Si
substrate with 215nm Si3N4 front anti-reflection coating (ARC) and
300nm Ag back reflection layer. Heights are not to scale.
H. Tian et al., Appl. Phys. Lett. (2017)
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel29
Direct Thermal Emission Measurement System
PM 1:
D=3”; EFL=4”
PM 2:
D=4”; EFL=4”
PM 3:
D=1.5”; EFL=2” FTIR
Chamber
Heater/Emitter
Cu Tubing
The sample is heated by the heater, and the emitted light is collected and guided by the Cu tube, transmitted
through a CaF2 window, reflected by three off-axis parabolic mirrors (PM 1, 2, and 3, Edmund Optics) to a Fourier
Transform InfraRed (FTIR) spectrometer with a mercury cadmium telluride detector and KBr beam splitter (Thermo
Fisher Nicolet 670).
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel30
H. Tian et al., Appl.
Phys. Lett. (2017)
Measurement (solid lines) and
simulation (dashed lines) of the
emissivity of selective absorbers
with (red lines) and without (black
lines) front coating at room
temperature. Measurements
performed by a Lambda 950
spectrophotometer with an
integrating sphere (Labsphere). The
thicknesses of Si3N4, Si and Ag are
215nm, 300 mm and 300nm
respectively.
Si
Si3N4
Ag
215 nm
300 nm
300 um
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel31
H. Tian et al., Appl.
Phys. Lett. (2017)
300 mm Si Experiment & Simulation at Room
Temperature
High spectral selectivity is observed at 468 ºC in both samples, with a cutoff wavelength of approximately
1.3 mm. Higher short-wavelength emittance is both predicted and observed for the structure with a Si3N4
AR coating
August 6, 201732
300 mm Si Experiment & Simulation at High Temperatures
with Si3N4 AR coating without Si3N4 AR coating
H. Tian et al., Appl. Phys. Lett. (2017)
SPIE Optics+Photonics, San Diego, CA - Peter Bermel
Thin Si film optimization targeted @ 550 °CEmissivity for selective absorbers
with different Si thicknesses.
Optimal Si3N4 thickness is used
for each curve which is 80 nm.
The temperature is set at
550℃ and the F-P interference
around the Mid-IR is smoothed
out for more clear comparison.
Less MWIR absorption is
experienced for thinner layers of
silicon because all samples are in
the intrinsic regime, and free
carrier absorption dominates.
300 um Si
20 um Si
5 um Si
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel33
H. Tian et al., Appl. Phys.
Lett. (2017)
Optimization Summary for 550 °C
Dependence of solar thermal transfer
efficiency 𝜂𝑡 for different Si thicknesses on
the concentration. The Si3N4 thickness is
fixed at 80nm, and the temperature is 550C.
Thinner layers of silicon experience less
reradiation; however layers which are too
thin have less absorption, which puts an
upper bound on 𝜂𝑡.
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel34
H. Tian et al., Appl. Phys. Lett. (2017)
Step 2: Reach Below-Ambient
Temperatures under Sunlight
Key tools:
RadCool
SPIE Optics+Photonics, San Diego, CA - Peter Bermel35 August 6, 2017
Radiative Cooling for Passive Thermal
Management
Photonic Crystal
Questions:
1. Any alternative coolers to PhCs?
2. What is the temperature reduction and
performance improvement by applying
radiative cooling to hybrid or STPV systems? Zhu, Linxiao et.al Proceedings of the National
Academy of Sciences 112.40 (2015): 12282-12287.
3 K microwave background
The sky transparency window
allows radiative cooling outdoors
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel36
Most PV cells experience heating from
sub-bandgap absorption
SPIE Optics+Photonics, San Diego, CA - Peter Bermel37 August 6, 2017
wavelength (nm)
Ab
so
rptivity
500 1000 1500 2000 25000
0.25
0.5
0.75
1
Po
we
r (W
/m2)
GaAs CIGS Si CdTe0
100
200
300
CIGS
CdTe
Si
GaAsAM 1.5G
Heat from
Sub-BG photons
(a)
(b)
wavelength (nm)
Ab
so
rptivity
500 1000 1500 2000 25000
0.25
0.5
0.75
1
Po
we
r (W
/m2)
GaAs CIGS Si CdTe0
100
200
300
CIGS
CdTe
Si
GaAsAM 1.5G
Heat from
Sub-BG photons
(a)
(b)In c-Si cells, degradation processes with activation energy of 0.85 eV are accelerated almost a factor of 2 for every 10
K temperature difference
X. Sun et al., IEEE J. Photovolt. (2017).
Radiative cooling on PV devices
Bare absorber
p-doped Si
AR
Al
p-doped Si
AR
Al
p-doped Si
AR
Al
Bare silica Silica PhC
Bare absorber +silica Bare absorber +silica
PhC
• Silica/silica PhC layer should at least preserve the solar absorption of
the absorber
• Silica/silica PhC layer is expected to enhance the thermal emittance at
the IR window
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel38
Radiative cooling on PV devices
Solar absorption of the three
structures
Emissivity spectra of the three
structures at the IR window
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel39
Experimental setup
Periodicity: 6 μm;
Depth: 10 μm;
The container allows control over convection
Zhu, Linxiao et.al Proceedings of the National Academy
of Sciences 112.40 (2015): 12282-12287.
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel40
Effects of radiative cooling
Without convection With convection
Zhu, Linxiao et.al Proceedings of the National Academy of Sciences 112.40 (2015): 12282-12287.
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel41
Benefits of radiative cooling extend across
many PV technologies and installations
SPIE Optics+Photonics, San Diego, CA - Peter Bermel42 August 6, 2017
Radiative Cooler
Glass
Solar Cell
Tedlar
Polymer
Polymer
Radiative Cooler
Enhanced
thermal radiation
T
PV (
K)
GaAs CIGS Si CdTe0
2
4
6
8
10 S. Cooling
R. Cooling
S.&R. Cooling
Concentration Factor
T
PV (
K)
1 2 3 4 50
10
20
30 S. Cooling
R. Cooling
S.&R. Cooling
Concentration Factor
TP
V (
K)
1 2 3 4 5
320
340
360
380
400 w/o cooling
(a)
(b)
X. Sun et al., IEEE J. Photovolt. (2017).
Methods
Radiative cooling – a passive technique that dissipates heat into remote
space via thermal radiation
Develop a simulation tool, RadCool, to model radiative cooling
Figure 3. Radiative cooling concept August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel43
Self-Consistent Modeling of Radiative
Cooling for Passive Thermal Management
SPIE Optics+Photonics, San Diego, CA - Peter Bermel44 August 6, 2017
Z. Zhou et al., SPIE Conf. Proc. 9973 (2016).
Radiative Cooling Reduces Temperature
and Improves Performance Substantially
SPIE Optics+Photonics, San Diego, CA - Peter Bermel45 August 6, 2017
Z. Zhou et al., SPIE Conf. Proc. (submitted).
Simulation tool - input
Heat load phase
Solar absorption power
𝑃𝑠𝑢𝑛 = 𝑐𝑜𝑛𝑐.∗ 𝐴 ∗ න0
∞
𝑑𝜆𝜀 𝜆 𝐼𝐴𝑀1.5(𝜆)
*Assuming incidence angle is 0
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel46
Simulation tool - input
Cooler phase
Thermal radiated power
𝑃𝑟𝑎𝑑
= න𝑑Ω𝑐𝑜𝑠𝜃න0
∞
𝑑𝜆𝐼𝐵𝐵𝜀(𝜆
*IBB is the spectral radiance of
a blackbody at temperature T
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel47
simulation tool - input
Environment Phase
Absorbed thermal radiation from the atmosphere
𝑃𝑎𝑡𝑚
= න𝑑Ω𝑐𝑜𝑠𝜃න0
∞
𝑑𝜆𝐼𝐵𝐵𝜀(𝜆, Ω)𝜀𝑎𝑡𝑚(𝜆, Ω)
Conductive Power
𝑃𝑐𝑜𝑑 = 𝐾 ∗ (𝑇 − 𝑇𝑐ℎ𝑎𝑚𝑏) Convective Power
𝑃𝑐𝑜𝑣 = 2 ∗ ℎ𝑐 ∗ 𝐴 ∗ (𝑇 − 𝑇𝑐ℎ𝑎𝑚𝑏)
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel48
Simulation tool - output
The steady-state temperature
T of the sample is determined
by:
𝑃𝑟𝑎𝑑 𝑇 − 𝑃𝑎𝑡𝑚 𝑇𝑎𝑚𝑏 − 𝑃𝑠𝑢𝑛+ 𝑃𝑐𝑜𝑑+𝑐𝑜𝑣 = 0
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel49
Experimental verification
Experimental data
Area ratio of the heat load and the cooler: 1
Cooling material: silicon wafer with soda-lime glass
Transmission spectrum: polyethylene film
Ambient temperature on the day of the experiment: ~290K
Experimental
Simulated
Tem
pera
ture
(K)
Time(min)August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel50
Conclusions/future work
RadCool successfully models radiative cooling system in a graphical interface
More experiments need to be done to confirm the generality of the system and
modeling approach
RadCool can be connected directly with the existing TPV model
The radiative cooling technique is not limited to TPV systems
Potential applications include solar cell cooling, infrared detectors,
and sensitive electronic devices that are used outdoors.
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel51
Step 3: Combine Hot and Cold Objects for
Maximum Efficiencies
Key tools:
TPXsim
TPVexpt
SPIE Optics+Photonics, San Diego, CA - Peter Bermel52 August 6, 2017
What Makes TPV Different from PV?
August 6, 201753 SPIE Optics+Photonics, San Diego, CA - Peter Bermel
PV cell efficiency TPV cell efficiency
TPV system efficiency
23% Demonstrated TPV Electric Generation
Efficiency with Spectral Control
B. Wernsman et al., IEEE Trans. Electron Dev. 51, 512 (2004)
Reflection spectrum for optical
filter and receiver
Efficiency in converting
radiation to electricity
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel54
Photon Recycling Can Greatly Reshape
High Temperature Thermal Emission
SPIE Optics+Photonics, San Diego, CA - Peter Bermel55 August 6, 2017
Ilic, Bermel et al., Nature
Nanotechnol. (2016)
TPV Efficiencies May Approach 52%* at
Reasonable Temperatures†
B. Wernsman et al., IEEE Trans.
Electron Dev. 51, 512 (2004)
*Using highly selective emitters shown above,
with MOVPE-grown GaSb TPV cells† World record η = 23% at 1050 °C
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel56
GaSb
0.7
Si
1.1
GaAs
1.4Material Choices:
Physics and Math
SPIE Optics+Photonics, San Diego, CA - Peter Bermel57 August 6, 2017
PV current Dark Current
Radiation Efficiency
Average Emissivity
TPXsim: A System-Level Modeling Tool
SPIE Optics+Photonics, San Diego, CA - Peter Bermel58 August 6, 2017
Unprecedented Efficiency of 35.4% is achievable for a filter band gap of 0.37 eV and PV band gap of 0.75 eV
Contour plot showing the combination of filter bandgap and PV
bandgap leading to maximum efficiency Or, just an emittance plot
Why is this meaningful?
Ongoing Research in Birck Nanotechnology
center will use these predictions to
experimentally fabricate and characterize
these structures
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel 59
TPXsim: A System-Level Modeling Tool
SPIE Optics+Photonics, San Diego, CA - Peter Bermel60 August 6, 2017
3
TPVexpt
• Based on TPVtest
• Considers complex rectangular geometries for heater,
emitter, and PV diode
• Considers non-idealities (e.g., series/shunt resistance)
• Phased GUI with overhauled “System” tab
SPIE Optics+Photonics, San Diego, CA - Peter Bermel61 August 6, 2017
4
View Factor/Geometry
• View factor: is the proportion of the radiation
which leaves surface A that strikes surface B
• View factor from cell to emitter for power calculations
View factor calculation for rectangle to rectangle in a
parallel plane. All boundaries are parallel or perpendicular
to x and ξ boundaries [2].
SPIE Optics+Photonics, San Diego, CA - Peter Bermel62 August 6, 2017
5
• Heater, emitter, and cell modeled as rectangles
• Emitter physically attached to heater
• Heater area broken up into four rectangles for
calculations
• Sum of rectangle view
factors is equal to heater
view factor
• Accounts for thickness
of emitter and
heater radiationMethod of calculating heater view factor. The back
rectangle represents the heater (dark blue). The
light blue rectangle represents the emitter.
Rectangle 1
Rec
tan
gle
3
Rec
tan
gle
2
Rectangle 4
View Factor Implementation
SPIE Optics+Photonics, San Diego, CA - Peter Bermel63 August 6, 2017
6
View Factor Effect on Output
Simulation 1
Simulation 2
Efficiency results. Simulation 1 results in
higher efficiency due to a greater view
factor (better alignment)
SPIE Optics+Photonics, San Diego, CA - Peter Bermel64 August 6, 2017
7
Shunt/Series Resistance
• Fill factor (FF) determines
the efficiency of PV cell
• Ideal cell has series
resistance of 0 Ω and
shunt of ∞ Ω
• Non-idealities decrease FF [6]Parasitic series and shunt
resistances in PV cell model [6].
SPIE Optics+Photonics, San Diego, CA - Peter Bermel65 August 6, 2017
8
Shunt/Series Effect on Output
Efficiency values with varying series and shunt
resistances. Default program values with
emitter-cell distance of 0.1 mm.
Ideal case (simulation 1)
Series only (simulation 2)
Shunt only (simulation 3)
Series and shunt (simulation 4)
Step 4: Improving Low-Bandgap
Photovoltaic Cells
Key tool(s):
ADEPT
MEEPPV
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel66
Drift-Diffusion Model Electrostatics (Poisson’s equation):
Charge conservation:
Current from drift & diffusion terms:
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel
D AV p n N Ne
1
1
J
J
n n
p p
nU
t q
pU
t q
S. Selberherr: "Analysis and Simulation of Semiconductor Devices“, Springer, 1984.67
( ) ( )
( ) ( )
n n n
p p p
dnJ qn x E x qD
dx
dnJ qp x E x qD
dx
m
m
Solar Cells: Ideal IV Characteristics
68PD = ITOTVD < 0
VD
ID
ITOT = I0 eqVD kBT -1( ) - ISC
-ISC
Pout = -ISCVD = 0
VOC
Pout = ITOTVOC = 0
Pout = ImpVmp = -ISCVOCFF
h =Pout
Pin=ISCVOCFF
Pin
ID = I0 eqVD kBT -1( )
Superposition
principle
Maximum
power
rectangle
August 6, 2017
SPIE Optics+Photonics, San Diego, CA - Peter Bermel
69
Maximum Short Circuit Current
Example: Silicon Eg = 1.1eV. Only photons with a wavelength < 1.12 mm
will be absorbed.
solar
spectrum
(AM1.5G)Pin = 100 mW cm2
l <hc
EG
JSC max= 44 mA cm2
August 6, 2017 SPIE Optics+Photonics, San Diego, CA - Peter
Bermel
70
Open-circuit Voltage and Efficiency
ITOT = I0 eqV /kBT -1( ) - ISC
I0 =1´10-12 A
VOC =kBT
qlnISC
I0
æ
èçö
ø÷
SPIE Optics+Photonics, San Diego, CA - Peter Bermel
VOC = 0.026 ln40 ´10-3
1´10-12
æ
èçö
ø÷= 0.63
ISC = 0.90´ 44 ´10-3 = 40 mA
h =Pout
Pin=ISCVOCFF
Pin
h =Pout
Pin=
40 ´ 0.63´ 0.8
100= 0.20
August 6, 2017
Example for silicon photovoltaics:
71
Increasing the Efficiency
SPIE Optics+Photonics, San Diego, CA - Peter
Bermel
h =Pout
Pin=ISCVOCFF
Pin
VOC =kBT
qlnISC
I0
æ
èçö
ø÷
I0 = qADn
WP
ni2
NA
æ
èçö
ø÷
1) Increase the short circuit current from 40 towards 44
2) Increase VOC (decrease I0)
August 6, 2017
Efficiency of Silicon Solar Cells (PERL Architecture)
Martin Green Group UNSW – Zhao et al., 1998 (25% at 1 sun)
72
370 - 400 mm
FF = 0.81
JSC = 41.5 mA/cm2 94%( )
VOC = 0.703 I0 = 0.075 ´10-12 A
August 6, 2017 SPIE Optics+Photonics, San Diego, CA - Peter
Bermel
73
JSC – VOC trade-off
1) Smaller bandgaps give higher
short circuit current
2) Larger bandgaps give higher
open-circuit voltage
3) For the given solar spectrum,
an optimum bandgap exists.
“Shockley-Queisser Limit”
SPIE Optics+Photonics, San Diego, CA - Peter
Bermel
August 6, 2017
ADEPT 2
Available on nanoHUB.org via:
https://nanohub.org/tools/adeptnpt/August 6, 2017
SPIE Optics+Photonics, San Diego, CA - Peter Bermel74
ADEPT: Input deck
Upon opening ADEPT 2, a blank input page will appear, awaiting your input file.
If upload/download does not work, one reason could be “pop-up” blocking by your internet browser.
Create a new input
Upload an input deck from your local computer
Download this input deck to your local computer
List of pre-loaded example input decks. Try them for
some quick simulation samples!
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel75
ADEPT: Running a simulation
Keep in mind that ADEPT 2 is FORTRAN 77 based. The format of certain input may cause unexpected error.
Please refer to “ADEPT 2 User Manual” for more information regarding how to write an ADEPT input deck.
This is your entire input file. You can edit it here.
Finally, click here to begin
simulation
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel76
ADEPT: While simulation is
running
In ADEPT, an entire simulation consists of two parts: ADEPT simulation and PLOTA output generation.
This window dynamically displays output.
Sometime, an error occurs and a notification will
be shown here.
This is a simulation progress bar. It shows
approximately how much simulation is done.
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel77
ADEPT: Output
Click here for a complete list of output plots
Click “input” to go back to input page. Worry not!
Your old simulation results will be saved until
you close ADEPT 2.0.
Click “clear” to clear out all output plots
You can review your old simulations results here
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel78
ADEPT: Output
Click on the plot and drag to “Zoom”
Download the plot as CSV or PDF image
Click “play” to look at this output quantity at different bias
“Zoom” reset
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel79
ADEPT: Output
Click on axis to format it
Bias sequence display option
Curve formatting
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel80
ADEPT: Output
Outputs include electrostatic (Poisson) solution:
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel81
ADEPT: Output Energy band diagram
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel82
ADEPT: Output Carrier concentrations:
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel83
ADEPT: Output And finally, realistic I-V curves:
August 6, 2017SPIE Optics+Photonics, San Diego, CA - Peter Bermel84
MEEPPV: User Interfaceshttps://nanohub.org/tools/meeppv
Upon opening MEEPPV,
a simulation option page
will appear, allowing
users to select between
using a graphical user
interface and uploading
a control file.
Click on the button
below to proceed to the
second page.
Input
parameters
that
describe
solar cell
simulationUpload
control file
similar to the
MEEP
interface85
Graphical User Interface
Click here to select
2D/3D solar cell for
simulation
You will be directed to this page if
the graphical user interface (first
option) is selected.
Users can input parameters that
describe the solar cell features
as well as the simulation domain.
86
Solar cell
schematic
MEEPPV Input
Feature tabs will appear when
the on/off button is turned on.
Click on each tab to input more
solar cell’s features.
Click on the on/off button to
include the features to the
solar cell for simulation.
The solar cell image shown on
the left changes with respect to
feature’s on/off button.
Click here to go back to the
first page.
Graphical User Interface
88
More input parameters under each feature tab
Finally, click here to begin
simulation
Graphical User Interface
89
Create new input
Download this control file to your local
computer
Upload a control file from your computer
Pre-loaded examples of control files. Try
them for some quick simulation samples.
This interface will appear if the second option (upload control
files) from the first page is selected.
If upload/download does not work, one reason could be “pop-
up” blocking by your internet browser.
Text-Based (Scheme) Interface
90
This is your entire
control file. You can
edit it here.
Note that the input file is written in Scheme language.
For more details and tutorial on writing control file with Scheme, please
refer to: http://ab-initio.mit.edu/wiki/index.php/Meep_Tutorial
Click here to begin
simulation
Text-Based (Scheme) Interface
91
This window
dynamically
displays output.
Sometimes, an
error occurs and
a notification will
be shown here.
Text-Based (Scheme) Interface
92
Output
Click here for different
output figures.
Click here to go back to
input page.
Click “clear” to clear all or one of the
simulation results.You can review
your old
simulation
results here.
93
Generating Graphics
Image sequence
display optionClick play to see the
animation of fields
propagating through the
solar cell94
MEEPPV Output
Downloading Data
Download current results to
your local computer
Click on axis to
format it
“Zoom”
97
MEEPPV: Post-processing in MATLAB
Summary
MEEPPV performs full-wave electromagnetics simulations of
photovoltaic devices
Two interfaces to control the input:
Graphical user interface – allows graphical feedback on
device design
Text-Based (Scheme) interface – allows greatest degree
of control, designed for experts
Output
Can generate graphics, including line plots (with
adjustable axes) and field distributions (either at a single
time, or as a movie)
Can download raw data as text or csv for further analysis
Any problems handled through nanoHUB help interface
99
Future Capabilities
Jupyter Notebooks
MATLAB-based version of ADEPT
100